SPLINE INTERPOLATION SYSTEM

Abstract

 This invention relates to a spline interpolation system which interpolates a given point by a cubic spline curve. A linear differential vector is determined from a predetermined number of points including a start point (P₁) and a cubic equation between the start point (P_I) and a next point (P₂) is determined from the coordinates value of predetermined points including the start point (P₁) and the end condition of the start point (P_i). Then, a spline curve between the start point (P,) and the next point (P₂) continuing from the former is determined. Next. a spline curve between points P₂ and P₃ is determined by adding the linear differential vector at a new point in place of the start point (P_i) and (P₂). In this manner, the cubic equations between points are obtained sequentially and a cubic spline curve is obtained. A spline curve free from a practical problem can be determined by sequentially reading points starting with the first without reading in advance the line of all the points at a time.

Description

Technical Field

[0001] The present invention relates to a spline interpolation method, and more particularly to, a spline interpolation method of effecting an approximate interpolation of a spline curve based on preset points among a sequence of given points.

Background Art

[0002] An interpolation method using a cubic spline curve is widely used as a method of subjecting a sequence of points to interpolation by using a smooth curve.

[0003] The prior art spline interpolation is effected by the following method:

(ia) all of the specified points of the sequence are received;

(iia) simultaneous equations containing all of the sequence point data are created and solved, to derive the first-derivative vector at each point. This is effected by deriving the inverse matrix of a matrix equation described later;

(iiia) under a condition that the first-derivative vector at the second point and the first-derivative vector at the starting point derived as described above are used, and that the positions of a given starting point and a second point are set, the following cubic equation connecting the starting point and the second point is derived, and the calculation is effected to derive the respective positions, thus effecting the interpolation while the parameter t of the equation is changed from 0 to 1,

(where P(t) denotes a vector representing the position, and K₄, K₃, K₂ and K₁ denote coefficient vectors); and

(iva) a cubic equation connecting the second and third points is derived in the same manner as described above.

[0004] Another method is effected as follows:

(ib) all of the specified points of the sequence are received;

(iib) simultaneous equations containing all of the sequence point data are created and solved, to derive the first-derivative vector at the starting point and the first-derivative vector at a second point, and derive the second-derivative vector at the starting point based on the above first-derivative vectors and the positions of the first and second points;

(iiib) a cubic equation, which is the same as described above and connects the starting point and the second point, is derived with the first-derivative vector and the second-derivative vector at the starting point derived as described above and used as the extreme point condition, and the calculation is effected to derive respective positions, thus effecting the interpolation while the parameter t of the equation is changed from 0 to 1; and

(ivb) a cubic equation connecting the second and third points is derived with the first- and second-derivative vectors of the second point in the cubic equation as described above, and the calculation then effected in the same manner as described above to derive the respective positions, thus effecting the interpolation while the parameter t of the equation is changed from 0 to 1.

[0005] In the above-described methods, the calculations are repeatedly effected for the third and fourth points, the fourth and fifth points, and so on, to effect the interpolation of the cubic equation for all of the specified sequential points so that the spline interpolation for all the sequential points can be effected.

[0006] In the above methods, however, all the specified sequential points must be received and the calculation therefor effected, and accordingly, the following problem arises:

(a) it is impossible to limitlessly increase the number of specified points;

(b) a very large amount of matrix calculations must be effected when the number of specified points is large, and thus a large memory capacity and a long calculation time are required;

(c) if the matrix calculation is not effected with a high precision, the spline curve may be excessively separated from the succeeding specified points (deviated or oscillated); and

(d) when the matrix calculation is effected on the programmer side, it is practically impossible to manually effect the calculation if the number of the specified points becomes equal to or larger than four, and thus an additional computing device becomes necessary.

Disclosure of the Invention

[0007] The present invention has been made in view of the above, and an object thereof is to provide a spline interpolation method of effecting an approximate interpolation of a spline curve based on preset points among a sequence of given points.

[0008] In this invention, to solve the problem described above, a spline interpolation method is provided in which given points are subjected to an interpolation using a cubic spline curve, and is characterized by deriving a first-derivative vector based on a preset number of points including a starting point, deriving a cubic equation between the starting point and a next point based on the coordinate values of the preset points including the starting point, the extreme point condition of the starting point, and the first-derivative vector, to derive a spline curve between the starting point and a point next to the starting point, and using a new point instead of the starting point and sequentially deriving a cubic equation between the points, to thereby derive a cubic spline curve.

[0009] First, a preset number of points starting from the starting point are used to derive a cubic spline curve, then a cubic spline curve is derived by using a new point instead of an initial point, and accordingly, a spline curve between points is sequentially derived to effect the spline interpolation process.

[0010] Therefore, the spline interpolation process can be effected without practical problems by sequentially receiving points in a forward direction without previously receiving all of the sequential points.

Brief Description of the Drawings

[0011] 

Fig. 1 is a diagram showing a spline curve;

Fig. 2 is a flowchart showing the process for deriving a spline curve in the present invention;

Fig. 3 is a block diagram of a numerical control device for effecting the present invention;

Fig. 4 is a diagram showing the construction of a matrix [A];

Fig. 5 is a diagram showing the construction of a matrix [B];

Fig. 6 is a diagram showing the construction of another matrix [A];

Fig. 7 is a diagram showing the construction of another matrix [B];

Fig. 8 is a diagram showing the general construction of a matrix _[A]-¹_;

Fig. 9 is a diagram showing the concrete construction of a matrix [A]-¹; and

Fig. 10 is a diagram showing the concrete construction of another matrix [A]-¹.

Best Mode of Carrying Out the Invention

[0012] An embodiment of the present invention will now be described with reference to the accompanying drawings. [Feature of this invention]

[0013] First, the feature of this invention is described. Attention is paid to the fact that, in the step of "deriving the first-derivative vector at each point based on all of the sequential points" in the conventional method (iia) of deriving a spline curve, a point among all of the sequential points nearer to a corresponding point at which the first-derivative vector is to be derived has a larger influence on the first-derivative vector, and the first-derivative vector at the corresponding point is approximately derived based on a sequence of points among all of the sequential points and nearer to the corresponding point, to create a cubic equation and effect the spline interpolation.

[0014] Further, a practical spline curve interpolation can be attained according to the following process, for example, set as the condition given in the starting point.

(a) Only a desired sequence of points is given and the second-derivative vector at the starting point is set to 0.

(b) The first-derivative vectors at the starting point and the desired sequential points are given. [Principle of this invention]

[0015] Next, the principle of the present invention is described.

[0016] Figure 1 shows a sequence of points subjected to a spline interpolation and a spline curve. Points P₁ to P_L indicate a sequence of points in which point P, is a starting point, point P_L is an end point, and points P_L+1 to P_L+m-1 are auxiliary points used for effecting an interpolation near the end point P_L. P₁(1) to P₄(1) indicate first-derivative vectors and SC is a spline curve.

(1) First-derivative vector

[0017] The first-derivative vector at each point of a cubic spline curve passing all of the sequential points P₁ to P_n (n points) is generally derived as follows.

[0018] A cubic equation constituting the spline curve is generally shown by the following equation:

where P(t) is a position vector which varies according to the parameter t, and K₄, K₃, K₂ and K₁ are coefficient vectors. That is, when an XYZ coordinate is used, P(t) = (x(t), y(t), z(t)) and K_i = (K_iX, K_iy, K_iz) (where i = 1, 2, 3 or 4). The parameter t is determined such that 0 ≤ t ≤ 1.

[0019] Accordingly, the first-derivative vector P⁽¹⁾ is expressed as follows:

[0020] At this time, the first-derivative vector [p⁽¹⁾] at each point can be expressed by the following equation:

[0021] Further, the matrix [P⁽¹⁾] of the first-derivative vector is a matrix of n rows x 1 column expressed by

and P₁⁽¹⁾ to P_n⁽¹⁾ are first-derivative vectors at P₁ to ^P_n'

[0022] At this time, it is generally known that the matrices [A] and [B] can be expressed as follows, according to the extreme point condition.

(i) In the case of the first extreme point condition:

(a) The second-derivative vector at the starting point (P₁) is 0.

(b) The second-derivative vector at the end point (P_n) is 0.

[0023] At this time, the matrix [A] is expressed by n rows x n columns, as shown in Fig. 4, and the matrix [B] is expressed by n rows x 1 column, as shown in Fig. 5.

(ii) In the case of the second extreme point condition:

(a) The first-derivative vector P₁(¹) at the starting point (P₁) is given.

(b) The second-derivative vector at the end point (P_n) is 0.

[0024] At this time, the matrix [A] is expressed by n rows x n columns, as shown in Fig. 6, and the matrix [B] is expressed by n rows x 1 column, as shown in Fig. 7.

[0025] From the equation (3), [P(¹)] can be derived by [p⁽¹⁾] = [A]^-1[B]. The matrix [A]^-1 is expressed by n rows x n columns, as shown in Fig. 8.

[0026] At this time, the absolute values of elements of the matrix [A]^-1 shown in Fig. 8 are large in the diagonal elements (a_k,k k = 1 to n) and become smaller in the elements which are farther away. Further, the matrix is a fixed matrix set by the extreme point condition.

[0027] Concretely, the matrix shown by an equation (5) of Fig. 9 is obtained when n is set at 50 in the extreme point condition 1. Further, the matrix shown by an equation (6) of Fig. 10 is obtained when n is set at 50 in the extreme point condition 2.

[Approximation of the first-derivative vector]

[0028] From the above-described theory, the first-derivative vector P₁⁽¹⁾ can be approximated as follows:

(the precision of the approximation is enhanced as k is made larger).

[0029] In this case, b_i is an i-th element of the matrix [B]. Based on this, P_i⁽¹⁾ can be derived without using all of the elements of [B], i.e., without using all of the sequential points.

[Formation of a cubic equation and interpolation]

[0030] If the first-derivative vector P_i⁽¹⁾ at P_i can be obtained, the following conditional equations can be derived by applying the points P_i and P_i ⁽¹⁾ to the equations (1) and (2):

[0031] (where Pi(0) and P_i(1) are values of Pi(t) when t = 0 and t = 1 where P(t) in the equation (1) is applied to P_i to P_i+1, and K_i4, K_i3, K_i2 and K_i1 are respective coefficients of t^3, t^2, t¹ and t⁰ in P_i(t). Further, P_i⁽¹⁾(0) and P_i⁽¹⁾(1) are values of the first-derivative vector when t = 0 and t = 1).

[0032] From the above four equations, four coefficients Ki₄, Ki_3' Ki₂ and Ki₁ can be obtained.

[0033] In this way, Pi(t) is sequentially derived and connected to create a spline curve.

[0034] When a cubic equation is derived by the above method, errors of P_i⁽¹⁾ and P_i⁽²⁾ with respect to the theoretical values thereof are not accumulated, and thus the errors can be suppressed to a negligible small value by adequately setting k in the equation (7). Therefore, a spline curve which poses no practical problems can be derived from the spline curve obtained by the present method, without enlarging the errors with respect to a spline curve obtained theoretically from a sequence of given points.

[Concrete example]

[0035] Next, a case wherein a spline curve is created by effecting the interpolation up to a point P_i and deriving the first-derivative vector P_i+1⁽¹⁾ at a next point is explained as a concrete example of the present method. [Creation of a first cubic equation]

[0036] First, a first cubic equation is created and subjected to interpolation.

(i) P₁ to P_1+m ((m+1) points; m is a constant and (m+1) is a number of buffers for receiving them) are received.

(ii) Assuming that P₁ is followed by a countless number of points, P₂⁽¹⁾ is derived from the equation (7).

(a) When the second-derivative vector P₁(2) is 0 at the starting point, P₂⁽¹⁾ is derived as follows by using the second row in the matrix of the equation (5), since the matrix obtained when assuming that P₁ is followed by a countless number of points is substantially equal to the matrix of the equation (5).

(b) Where the first-derivative vector P₁⁽¹⁾ is given at the starting point, P₂⁽¹⁾ is derived as follows by using the second row in the matrix of the equation (6) since the matrix obtained when assuming that P₁ is followed by a countless number of points is substantially equal to the matrix of the equation (6).

[0037] The following condition (a) or (b) is used according to the extreme point condition:

(a) Where the second-derivative vector P₁⁽²⁾ is 0 at the starting point,

(b) Where the first-derivative vector P₁⁽¹⁾ is given at the starting point,

[0038] Coefficients K₄, K₃, K₂ and K₁ are derived from the four conditional equations (12), (13), (14) and (15), to create a cubic equation, and the cubic equation is subjected to interpolation.

[Creation of second and later cubic equations]

[0039] Next, a second or later cubic equation (i-th; i>=2) is created and is subjected to interpolation.

(i) P_i+m is received (a state is set up in which (m+1) points P_i to P_i+m are received).

(ii) Assuming that a spline curve on which P_i is followed by a countless number of points is given, an approximate tangential vector P_i+1⁽¹⁾ at the second point on the curve is derived by using the equation (7). pi+1⁽¹⁾ is derived as follows, using the second row in the matrix of the equation (6), since the matrix obtained when assuming that P₁ is followed by a countless number of points is substantially equal to the matrix of the equation (6).

(P_i⁽¹⁾ has been derived by the spline curve up to p_i)

[0040] Coefficients K₄, K₃, K₂ and K₁ are derived from the above conditional equations to create a cubic equation, and the cubic equation is subjected to interpolation.

[0041] (3) When the end point P_L is reached, the process (2) is repeatedly effected to sequentially create cubic equations between points P_L-m to P_L while auxiliary points P_L, P_L+1' ---_' P_L+m-1 are created.

[0042] In this case,

[Flowchart]

[0043] The processing flowchart of the present invention is shown in Fig. 2. In Fig. 2, each numeral following S denotes a step number.

[S1] Set i to 1.

[S2] Receive points P_i to P_i+m. Make (m+1) points by creating auxiliary points when a point to be received has exceeded the end point.

[S3] Execute S4 when i = 1, i.e., the starting point is specified; and execute S5 when a point other than the starting point is specified.

[S4] Create a cubic equation under a condition including the starting point.

[S5] Create a cubic equation under a condition excluding the starting point.

[S6] Derive coefficients of the created cubic equation and effect the interpolation between points by using the cubic equation.

[S7] Increment i by 1 and effect the interpolation for a next cubic equation between points.

[S8] Determine whether or not P_i is an end point, and if it is not the end point, the process returns to S2 and the interpolation is continuously effected.

[Construction of hardware]

[0044] Figure 3 shows the construction of the hardware of a numerical control device (CNC) for effecting the present invention. In Fig. 3, 11 denotes a processor for global control (CPU), 12 denotes a ROM in which a control program and the like used for deriving a spline curve are stored, 13 denotes a RAM for storing interim data and the like of a point on the spline curve, and 14 denotes a nonvolatile memory in which the derived spline curve, machining program, parameters and the like are stored. Numeral 14a denotes a machining program.

[0045] Numeral 15 denotes a PMC (programmable machine controller) which receives instructions such as M-function and T-function, converts them into signals for controlling the machine tool, and outputs the same. The M-function and the like are processed according to a sequence program 15a and then supplied from an input/output circuit to a control circuit on the machine side. Numeral 16 denotes a display control circuit which converts a digital signal into a display signal; 16a denotes a display unit, which can be a CRT, liquid crystal display device or the like; and 17 denotes a keyboard used for inputting various data.

[0046] Numeral 18 denotes a position control circuit for controlling a servomotor; 19 denotes a servo amplifier for controlling the speed of the servomotor; 20 denotes the servomotor; 21 denotes a tachometer generator for effecting the speed feedback; and 22 denotes a position detector, which can be a pulse coder, optical scale or the like. The above elements must be provided in a number corresponding to the number of axes, but in this case, the elements for only one axis are shown.

[0047] Numeral 23 denotes an input/output circuit for transferring digital signals with respect to the external unit, and a tool select signal (T-signal) for controlling the replacement of tools is output from the circuit to the machine-side control circuit. Numeral 24 denotes a manual pulse generator for moving each axis in a digital manner.

[0048] In the above construction, a single processor is used, but it is also possible to use a processor for global control and a processor for deriving a spline curve, thus constructing a multi-processor. Further, it is possible to add a co-processor or the like to enhance the operation speed.

[0049] Further, an automatic programming device other than the numerical control device can be applied to carry out the spline interpolation method of this invention.

[0050] As described above, according to the present invention, since a cubic spline curve between points is sequentially derived based on a preset number of points among a sequence of given points, a practical spline curve can be derived in a short period of time even when the number of points to be interpolated becomes larger.

[0051] Further, the extreme point condition is set with the second-derivative vector set at 0, so that the interpolation of the spline curve can be effected without setting the condition of the first-derivative vector at the extreme point or the starting point.

[0052] In addition, a smooth spline curve can be obtained near the end point by providing auxiliary points after the end point.

Claims

1. A spline interpolation method for subjecting given points to interpolation by using a cubic spline curve, characterized by

deriving a first-derivative vector from a preset number of points including a starting point;

deriving a cubic equation between said starting point and a next point based on coordinate values of the preset number of points including said starting point, the extreme point condition of said starting point and said first-derivative vector to derive a spline curve between said starting point and a point next to said starting point; and

using a new point instead of said starting point and sequentially deriving a cubic equation between points, to thereby derive a cubic spline curve.

2. A spline interpolation method according to claim 1, characterized in that the spline curve is subjected to interpolation in said extreme point condition in which second-derivative vectors at said starting and end points are set at 0.

3. A spline interpolation method according to claim 1, characterized in that the spline curve is subjected to interpolation in said extreme point condition in which the first-derivative vector at the starting point is given and a second-derivative vector at an end point is set at 0.

4. A spline interpolation method according to claim 1, characterized in that a spline curve near an end point is derived by using auxiliary points after the end point.

Drawing